Several systems or functions are believed to be pro- grammed in utero (Barker, 1998; 2001). These are only briefly reviewed here, as our focus is CVD and risk markers.
3.7.1 Infection, immunity and autoimmune diseases
Presently, only a few studies link adult immune func- tion to birth size. One report links disproportionate fetal growth (acute IUGR) with altered IgE concentra- tions in adults (Godfrey, Barker & Osmond, 1994). A few studies have observed an association between birth weight and autoimmune diseases. LBW at term was found to be inversely related to autoimmune thyroid disease (Phillips et al., 1993; Barker, 1998). It has also been suggested that type-1 diabetes might be pro- grammed in utero, based on the observed association of risk in children with birth weight in Sweden, after controlling for maternal age, parity, diabetes, and smoking (Dahlquist, Bennich & Källen, 1996). The ob- served association, however, was positive, while it is negative for type-2 diabetes. The case-control study of children under 14 years of age and diagnosed for dia- betes showed that children born large for gestational age (z-score of ≥2) were at significantly higher risk, while those born small (z-score <2) were at lower risk. Based on cohort studies in the Gambia and in view of animal evidence, fetal programming of immunity was hypothesized (Moore et al., 1997; Prentice et al., 1999). In this country with yearly seasonal patterns of food shortage, birth weights were found to be 200–300 g lower in the hungry season. Impaired fetal growth causing high rates of LBW in the hungry season is nu- tritionally mediated, since the incidence of LBW was reduced by supplementary feeding of mothers (Ceesay et al., 1997). Three villages have kept accurate vital sta- tistics since 1949, which made it possible to study sur- vival and causes of death, using season of birth as a proxy for early environmental insults. The cohort con- sisted of more than 3, 000 individuals. Similar mortal- ity patterns were observed up to puberty irrespective of season of birth. There was a marked increase there- after, with a risk ratio of 3.7 for deaths after 14.5 years, and of 10.3 for deaths after 25 years, in people born during the hungry season. Causes of death were pri-
marily infection or infection-related, not chronic dis- eases. It was therefore suggested that early life events caused permanent damage to the immune system. In support of this hypothesis, a longitudinal study in the Philippines by McDade et al. (2001) found that im- paired fetal growth as reflected in small-for-gestational- age (SGA) at birth was associated with reduced plasma thymopoietin concentration at adolescence (103 sub- jects), after controlling for potential confounding vari- ables. It is known that thymic hormones play an important role in T-cell development and function, and therefore, altered thymus development during fetal life could programme impaired immunity in later life as well as affecting survival in early years. In addition to seasonal undernutrition, other potential initial insults in the Gambian study include exposure to aflatoxin, prenatal malaria, or postnatal viral infection. As re- viewed by Prentice (1998), there is abundant evidence of sustained effects of fetal undernutrition on immu- nological function in animals. Zinc depletion during gestation, in particular, has a profound effect on the development of lymphoid organs, especially the thy- mus and spleen, and later repletion is ineffective. Intergenerational effects of maternal zinc deficiency have even been reported in the ’70s. Similar effects may result from protein-energy or other micronutrient de- ficiencies, but zinc is known to play a prominent role in the modulation of immune function.
The reported inverse association of birth weight with asthma is not unrelated to possible early programming of immunity. As reviewed by Shaheen et al. (1999), some studies have found a relation between birth weight and asthma in children, although most have not. Two large studies of male conscripts have found an association (Seidman et al., 1991b; Braback & Hedberg, 1998). Using a British cohort born in 1970, Shaheen et al. (1999) investigated the relationship between birth weight and childhood (age 10) and adult anthropom- etry (age 26) and adult asthma/wheezing in adults. Many confounding factors were controlled for, includ- ing maternal age, smoking history in pregnancy, dura- tion of breastfeeding, socioeconomic status of the father, and number of children in the household. It was found that among respondents (age 26), asthma and wheezing fell with increasing birth weight, and this effect was even stronger when gestational age was con- trolled for. The odds ratio was 3.3 (CI: 1.31–7.46) for those with birth weights lower than 2 kg compared with birth weights between 3–3.5 kg. Asthma and wheezing also increased as BMI increased, particularly so in the last quintile. Controlling for birth weight increased the estimated effect of BMI. The relationship was also stronger for women, who were more affected by asthma
than men. Impaired fetal growth of airways, leading to small airway size, could be responsible for the associa- tion, as described by Harding et al. (2000).
3.7.2 Breast cancer
It has been postulated that breast cancer may originate in utero. Available evidence, however, is in the oppo- site direction of that of CVD, with higher birth weights associated with higher odds ratio for breast cancer (Signorello & Trichopoulos, 1998). However, current evidence for perinatal influences on breast cancer risk is less consistent than the evidence for perinatal influ- ences on CVD.
The hypothesis of early life modulation of breast cancer risk is supported by immigrant studies. The rate of breast cancer in Japanese migrants rises to the level observed in white Americans only after two or more generations, whereas colorectal cancer becomes simi- lar to the host population within a few decades of mi- gration. This suggests that the latter is influenced more by lifestyle factors than the former (Ekbom et al., 1992; Ziegler et al., 1993).
An intrauterine origin could be linked to estrogens or other hormones, or to other pregnancy-related proc- esses. Fetal growth is associated with estrogens, but primarily at the extreme ends of the range. Whether the fetus is exposed to any of the estrogens in the moth- er’s circulation is unclear, however. Studies have either found no association of birth weight with breast can- cer, or a positive trend. Sanderson et al. (1996) found a positive association, but only in premenopausal women, which lends support to the hypothesis linking breast cancer with pregnancy estrogens. In the large prospective cohort studies on American nurses, a nested case-control study (Michels et al., 1996) revealed that for women with a birth weight of ≥4 kg, there was a graded reduction of risk which reached 45% in those weighing 2.5–2.9 kg at birth. In Sweden, Hübinette et al. (2001a) reported in a case-control study on breast cancer in sister twins (96 pairs). Case pairs were com- pared with external age-matched twin pairs. Healthy sister twins were the internal control group. Gestational age >40 weeks considerably increased the risk com- pared to twins of gestational age <32 weeks (OR: 8.4; CI: 1.3–54.4). It was also found that cases had signifi- cantly higher birth weight and PI at birth. The risk in- creased with increasing birth weight, but the trend was not significant. Another study in Swedish women (Andersson et al., 2001) revealed that the risk of all site cancer and non-hormonal type of cancer was also in- creased with increasing birth weight and birth length, even after controlling for potentially confounding vari-
ables. This suggests that size at birth may influence cancer sites other than hormone-related sites like the breast. However, more research is required to elucidate this relationship.
3.7.3 Polycystic ovary syndrome
Polycystic ovaries in adult life have also been linked with fetal growth and length of gestation (Cresswell et al., 1997). Placental failure associated with postmatu- rity may expose the hypothalamus to increased con- centrations of androgens and estrogens, and reset its responses to them. There could be an altered hypothalamic-pituitary “set point” for luteinizing hor- mone (LH) release, and this may programme one type of polycystic ovaries (Cresswell et al., 1997). Another form may also be programmed in utero, and is associ- ated with high birth weight, high maternal weight, and high current BMI. It is interesting to note that women with polycystic ovaries are at higher risk of diabetes and CVD after menopause, and that in such women, hyperinsulinemia and insulin resistance tend to worsen spontaneously (Pasquali et al., 1999).
3.7.4 Other diseases suspected of fetal programming
Among the several conditions that are postulated as being programmed in utero, renal disease among Amerindians (Hoy et al., 1999), and osteoporosis mark- ers in Caucasian populations (Cooper et al., 1997; Jones, Riley & Dwyer, 1999) have been studied. How- ever, as seen in Appendix A, very limited evidence ex- ists to date. Reports of psychological and mental repercussions of early life experience in the literature are dated (Stein et al., 1975), but there has been a recent upsurge of interest, with data suggesting that behavioural development and mental disorders may have a link with prenatal growth. In a population-based UK cohort study of people born in 1946, birthweight was significantly and positively associated with cogni- tive ability at age 8 and older, although the association was less strong than at age 8 (Richards et al., 2001). The association was strengthened when controlling for socioeconomic status. It was suggested that insulin dep- rivation, or altered action in the growing brain may be involved (Wickelgren, 1998).